In Vitro Phosphorylation of GFAP and Desmin by Aurora-B
We reported that vimentin is phosphorylated at Ser-72 by an unidentified kinase activity at the cleavage furrow in the late mitotic phase (Yasui et al., 2001
). Because Aurora-B is localized at the spindle midzone and the equatorial cortex, it was considered to be a candidate for the novel protein kinase. We therefore determined if Aurora-B phosphorylates type III intermediate filament (IF) proteins, including GFAP and desmin, in vitro. As shown in Figure , A and B, these filament proteins were indeed phosphorylated by Aurora-B, but not by Aurora-B(K/R), which is a kinase-negative version with a mutation at the ATP-binding site. The phosphorylation level of GFAP increased in a time-dependent manner up to 1.0 mol of phosphate/mol of protein (Figure C), whereas the level of desmin reached 1.5 mol of phosphate/mol of protein (Figure D). We then examined effects of the phosphorylation on the filament forming potential of GFAP and desmin. Soluble GFAP or desmin was preincubated with Aurora-B or the K/R mutant for 1 h, and the samples were incubated under conditions of polymerization. Analyses of these samples by centrifugation (Figure E) and by electron microscopy (Figure F) revealed that the phosphorylation of GFAP or desmin by Aurora-B dramatically inhibited filament formation. Thus, phosphorylation of GFAP and desmin by Aurora-B affects the state of polymerization of filaments in vitro.
Figure 1 Phosphorylation of GFAP and desmin by Aurora-B and the effect of phosphorylation on filament formation. GFAP (A) and desmin (B) were treated with Aurora-B (WT) or Aurora-B (K/R) and analyzed using SDS-PAGE followed by autoradiography. (C and D) Time course (more ...)
Identification of In Vitro Phosphorylation Sites on GFAP and Desmin by Aurora-B
To identify the sites on GFAP phosphorylated by Aurora-B, GFAP was first exposed to Aurora-B in the presence of [γ-32P]ATP. Phosphorylated GFAP was treated with lysyl-endopeptidase and subjected to reverse-phase HPLC, and the resultant peptide was digested with trypsin and again applied to a reverse-phase HPLC column. As shown in Figure A, three major radioactive peptides, G1 to G3, were obtained, the G1 peptide was phosphorylated at a threonine residue, and G2 and G3 peptides were phosphorylated at serine residues, as determined by phosphoamino acid analysis (Figure C). The phosphoserine-containing peptides were then sequenced after ethanethiol treatment which specifically converts phosphoserine into S-ethylcysteine. The generation of S-ethylcysteine at a particular Edman degradation cycle where serine is predicted provides a definitive way to locate the phosphoserine residue(s) on each peptide. The lack of generation of S-ethylcysteine indicates that phosphoserine is located in the amino-terminal serine residue as there is no conversion of the amino-terminal phosphoserine to S-ethylcysteine. Indeed, phosphates of G1, G2, and G3 peptides were found to be located on Thr-7, Ser-38, and Ser-13 on GFAP, respectively (Table ). We also examined the sites phosphorylated by Aurora-B on desmin using the same methods described above. A peptide derived from phosphorylated desmin digested with lysyl-endopeptidase was treated with trypsin and subjected to reverse-phase HPLC. The three major radioactive peptides, D1 to D3, were obtained (Figure B), and D1 and D2 peptides were phosphorylated at serine residues, whereas the D3 peptide was phosphorylated at a threonine residue, as determined by phosphoamino acid analysis (Figure D). Phosphates of D1, D2, and D3 peptides were found to be located on Ser-11, Ser-59, and Thr-16, respectively (Table ).
Figure 2 Analysis of trypsin-digested fragments of phosphorylated GFAP and desmin. (A and B) The radioactive single peptide of GFAP (A) and desmin (B) was digested with trypsin and fractionated by reverse-phase HPLC. The radioactivity of each fraction was measured (more ...)
Amino acid sequence of 32P-labeled phosphopeptides of GFAP and desmin by Aurora-B
Effects of Mutation of Desmin at Aurora-B and Rho-kinase Phosphorylation Sites on Filament Formation
To examine effects of phosphorylation on the filament organization of desmin, we constructed a set of desmin mutants in which in vitro phosphorylation sites by Aurora-B (Ser-11, Thr-16, and Ser-59), Rho-kinase (Thr-16, Thr-75, and Thr-76), or both were changed to Asp. These mutants are assumed to mimic desmin phosphorylated by Aurora-B, Rho-kinase, or both. When wild-type desmin or these mutants were transiently expressed in T24 cells that do not express any type III IFs, wild-type desmin showed a filamentous pattern (Figure , A and B). Desmin with mutations in Rho-kinase or Aurora-B phosphorylation sites was no longer filamentous rather was diffusely distributed in the cytoplasm (Figure , C–F). Desmin with mutations in Rho-kinase and Aurora-B phosphorylation sites showed a markedly diffuse localization (Figure , G and H). These findings suggest that the in vitro desmin phosphorylation at sites of Aurora-B and/or Rho-kinase induce disassembly of the filaments in vivo. The same phenotype was observed when a GFAP mutant in which in vitro Aurora-B phosphorylation sites were changed to Asp was expressed in cells (unpublished data). It is notable that 100% of the cells expressing the desmin or GFAP mutants showed diffuse localization of the expressed IFs.
Figure 3 Immunofluorescence staining of T24 cells expressing desmin and the mutants. T24 cells were transfected with pDR2-desmin (A and B) or pDR2-mutated desmin in which phosphorylation sites by Rho-kinase (C and D), Aurora-B (E and F), or both (G and H) are (more ...)
Comparison of In Vitro Phosphorylation Sites by Aurora-B to Rho-kinase and Production of the Site- and Phosphorylation State-specific Antibodies for Desmin
In a previous study, we found that Rho-kinase phosphorylates GFAP at Thr-7, Ser-13, and Ser-38 in vitro (Figure A; Kosako et al., 1997
), and the phosphorylation is observed at the cleavage furrow during cytokinesis in vivo (Nishizawa et al., 1991
; Matsuoka et al., 1992
; Kosako et al., 1997
; Yasui et al., 1998
). In the present study we found the same phosphorylation sites to be those for Aurora-B in vitro (Table ). When we made a Western blot analysis with TMG7, YC10, KT13, or KT34, which reacts with GFAP phosphorylated at Thr-7, Ser-8, Ser-13, or Ser-38, respectively (Nishizawa et al., 1991
; Matsuoka et al., 1992
; Kosako et al., 1997
; Yasui et al., 1998
), we confirmed that Aurora-B phosphorylates GFAP at Thr-7, Ser-13, and Ser-38 but not at Ser-8, whereas PKA phosphorylated all four residues (Figure B).
Figure 4 Phosphorylation site maps of GFAP and desmin by Aurora-B and Rho-kinase and characterization of antibodies by Western blotting. (A and C) Location of the sites on GFAP (A) and desmin (C) phosphorylated by Aurora-B and Rho-kinase is shown. The phosphorylation (more ...)
Rho-kinase was found to phosphorylate desmin at Thr-16, Thr-75, and Thr-76 (Figure C; Inada et al., 1999
). In the present study, we found Aurora-B phosphorylates desmin at Ser-11, Thr-16, and Ser-59, in vitro. Because Ser-59 of desmin is a phosphorylation site specific to Aurora-B in vitro, the residue might serve as a pertinent indicator to study in vivo desmin phosphorylation by Aurora-B. We next prepared rabbit polyclonal antibodies (referred to as α-PD11 and α-PD59), raised against the synthetic phosphopeptides, PD11 and PD59. In Figure , D and E, the specificity of α-PD11 and α-PD59 was examined and compared with that of other antiphosphodesmin antibodies such as α-PD16, α-PD75, or α-PD76, using Western blotting. Although α-PD59 specifically reacted with desmin phosphorylated by Aurora-B, but not that by Rho-kinase and PKA, Ser-11 as well as Thr-16 and Thr-75 of desmin was phosphorylated by PKA in vitro (Figure D). As shown in Figure E, the immunoreactivity of α-PD11 and α-PD59 for desmin phosphorylated by Aurora-B was neutralized by preincubation with PD11 and PD59, respectively, but not with phosphopeptides PD16, PD75, and PD76; thus, the specificity of the two antibodies was confirmed.
Phosphorylation of Desmin at Ser-11 and Ser-59 by Ectopic Expression of Active Aurora-B in COS-7 Cells
The inner centromere protein (INCENP) was reported to interact with Aurora-B and be phosphorylated by the kinase, after which Aurora-B kinase activity was increased, in vitro (Kaitna et al., 2000
; Bishop and Schumacher, 2002
). On the other hand, RNA interference (RNAi) study using Drosophila
S2 cells revealed that INCENP is essential for Aurora-B–mediated phosphorylation of histone H3 at Ser10 (Adams et al., 2001b
). In the present study, we observed the phosphorylation of histone H3 at Ser-10 in interphase when Aurora-B is coexpressed with INCENP in COS-7 cells (Figure , A–C). The findings demonstrated that Aurora-B is activated through interactions with INCENP in vivo as well as in vitro. Under these conditions, activated Aurora-B also induced phosphorylation at Ser-28 of histone H3 in interphase cells (Figure D). We next asked if Aurora-B phosphorylates GFAP or desmin at the sites identified in vitro, using COS-7 cells and transient transfection analyses. As shown in Figure E, phosphorylation of GFAP at Thr-7, Ser-13, and Ser-38 but not Ser-8 was detected in interphase COS-7 cells expressing GFAP with Aurora-B and INCENP. The same phosphorylation pattern was observed when the catalytic domain of Rho-kinase (Rho-K-CAT) was used instead of Aurora-B and INCENP (Figure E). Next, we determined if desmin is phosphorylated at Ser-11, Thr-16, and Ser-59 by activated Aurora-B in interphase when Aurora-B and INCENP are overexpressed with desmin in COS-7 cells. COS-7 cells ectopically expressing desmin, Aurora-B and INCENP were immunostained with 9E10 and an antiphosphodesmin antibody: α-PD11, α-PD16, α-PD59, α-PD75, or α-PD76. As shown in Figure F, phosphorylation of overexpressed desmin at Ser-11, Thr-16, and Ser-59 was observed in cells coexpressing Aurora-B and INCENP, whereas that of desmin-Thr-75 and Thr-76 was not detected. We also examined the phosphorylation of desmin at Thr-16, Thr-75, and Thr-76 in interphase COS-7 cells ectopically expressing Rho-K-CAT. Phosphorylation of desmin at Thr-16, Thr-75, and Thr-76 was observed in interphase cells expressing Rho-K-CAT (Figure F). The phosphorylation of desmin at Ser-11 and Ser-59 was not observed in cells expressing Rho-K-CAT. Therefore, Ser-11 and Ser-59 could be the sites phosphorylated by active Aurora-B in cells.
Figure 5 Phosphorylation of histone H3, GFAP, and desmin in COS-7 cells expressing Aurora-B activated by INCENP or active Rho-kinase. (A–D) COS-7 cells were transfected with pcDNA-Myc-Aurora-B (A), pCMV-HA-INCENP (B), or both (C and D) and double-stained (more ...)
Specific Phosphorylation of Desmin at Ser-11 and Ser-59 at the Cleavage Furrow during Cytokinesis
Based on the foregoing biochemical and immunocytochemical observations that desmin-Ser-11, Thr-16, and Ser-59 are sites phosphorylated by Aurora-B in vitro and that desmin-Thr-16, Thr-75, and Thr-76 are phosphorylation sites by Rho-kinase, the spatial and temporal distributions of the five phosphorylation sites in Saos-2 human osteosarcoma cells were analyzed, using α-PD11, α-PD16, α-PD59, α-PD75, and α-PD76. Under the conditions used, all immunoreactivity of these antiphosphodesmin antibodies was detected only in late mitotic cells and specifically at the cleavage furrow (Figure A), but not in interphase cells or in early mitotic cells such as prometaphase or metaphase (unpublished data). When desmin was stained with an α-desmin antibody, which reacts with both phosphorylated and unphosphorylated desmin, the filamentous structure was observed in mitotic daughter cells (Figure A) and in interphase cells (unpublished data). Rho-kinase accumulated at the cleavage furrow and Aurora-B was enriched at the spindle midzone during the cytokinesis (Figure B). These observations may suggest that desmin filaments are phosphorylated by Aurora-B as well as Rho-kinase, at the cleavage furrow during cytokinesis.
Figure 6 Localization of phosphodesmin, Aurora-B, and Rho-kinase. (A) Fluorescent photomicrographs of mitotic Saos-2 cells stained with α-PD11, α-PD16, α-PD59, α-PD75, α-PD76, or α-desmin (green). Chromosomes were (more ...)
Effects on Cytokinesis of Desmin Mutants at Aurora-B and/or Rho-kinase Phosphorylation Sites
In the next set of experiments, we determined if a desmin mutant at in vitro phosphorylation sites by Aurora-B or Rho-kinase induces IF-bridge formation. We constructed three desmin mutants, in which sites phosphorylated in vitro by Rho-kinase, Aurora-B, or both are changed to Ala or Gly. When the mutants were transiently expressed in T24 cells, the mutant desmin filaments failed to segregate into daughter cells; rather they formed an unseparated bridge-like structure between them (Figure A and unpublished data). When we evaluated the percentage of postmitotic IF-bridge–forming cells, ~7% and ~23% of the cells expressing mutated desmin at Rho-kinase sites and at Aurora-B sites, respectively, formed a desmin-bridge structure (Figure B). Mutations at both Rho-kinase and Aurora-B phosphorylation sites also showed effects, and ~35% of the transfected cells had the desmin-bridge (Figure B). On the contrary, expression patterns of the mutants in interphase cells were indistinguishable from those of wild-type desmin (Figure C). Although cells expressing the mutants had a normal morphology at prometaphase, metaphase, anaphase, and telophase (unpublished data), they did have a striking phenotype after passing through telophase.
Figure 7 Effects of mutations in desmin at Aurora-B and/or Rho-kinase phosphorylation sites on cytokinesis. (A) IF bridge-like structures in T24 cells expressing a desmin mutant at phosphorylation sites by Rho-kinase and Aurora-B; Thr-16, Ser-59, Thr-75, and Thr-76 (more ...)